This paper examines the efficiency market hypothesis for the art market using a novel nonlinear quantile‐based unit root test while accounting for sharp shifts and smooth breaks in the data. We use quarterly data which covers 1998:1 and 2015:1. Our analysis is based on 15 art price indices: Contemporary, Drawings, France, Global index (Euro), Global index (USD), Modern art, Nineteenth century, Old Masters, Paintings, Photographies, Postwar, Prints, Sculptures, UK and US. We find evidence of structural shifts and nonlinearity in the art indices. We cannot reject the null of unit root and/or stationarity in the art series based on the conventional linear unit root tests and quantile‐based test that did not account for structural breaks. However, when we use the same methods but accounting for sharp shifts and smooth breaks, we are able to reject the unit root null for each of the art indices. Further we find evidence of asymmetric behaviour in some of the indices: two global indices, Paintings, Prints, Sculptures, Modern art and Postwar, where unit root exist at some quantiles but not at others. Overall, our result suggest that the art market is inefficient. We provide some practical and policy implications of our findings.